AI for humanity

I'm very happy they list "Opening up the black boxes of AI" as a main point.

While we have very powerful classification devices able to navigate the subtle hidden topology of a problem, this is still correlation.

To me Explainable ML will be key to go from correlation to causality. Explanability will help by pointing where to investigate and hypothesize and test for some causes. It could be the missing link between ML and Judea Pearl's causality theory.

DARPA has a program on it [1], NIPS a symposium [2]. This is a significant dimension that is growing in the field.

[1] https://www.darpa.mil/program/explainable-artificial-intelli...

[2] http://interpretable.ml/

sama is in a panel at about 2:05 into the video, themed on global AI market competition.